Teams of Rivals: Learning About a Cabinet and Its Shadow
Total Page:16
File Type:pdf, Size:1020Kb
Political Science and Political Economy Working Paper Department of Government London School of Economics No. 4/2012 Teams of Rivals: Learning about a Cabinet and its Shadow Torun Dewan (LSE) Rafael Hortala-Vallve (LSE) Teams of Rivals: Learning about a Cabinet and its Shadow∗ Torun Dewan London School of Economics and Political Science [email protected] Rafael Hortala-Vallve London School of Economics and Political Science [email protected] Abstract. We explore electoral accountability in a model in which an incumbent team chooses between a \safe" option with a known payoff and a \reform" policy that yields higher or lower payoffs depending on whether the politician implementing it is competent or not. The first best outcome allows for learning about politicians' type, but is subject to the free-rider problem: if team members are moderately competent, on average, they prefer others to implement the reforms. A cabinet provides an institutional remedy by allowing politicians to obtain individual rents from office and making policy choice subject to unanimous consent. Although free-riding is alleviated, outcomes with a cabinet are qualitatively similar to those in its absence. We relate this to a hold-up problem: politicians revealed as incompetent veto the implementation of reform. Surprisingly, the problem is most severe when such politicians can be removed from office. Electoral competition between rival teams, as in that between a cabinet and a shadow cabinet, restores learning; but at moderately high levels of competence, relative to the efficient benchmark, such competition leads to over-investment in risky reforms. This contrasts with the case where executive decision-making is by an individual{then electoral competition leads to first best outcomes. ∗For helpful comments we thank Gilat Levy, Hande Mutlu-Eren, participants at the \Priorat Workshop on Formal Models in Political Science 2012", and audience members at the Annual Meetings of the European Political Science Association, Berlin, 2012. 1 2 1. Introduction In many situations of political interest rival teams of politicians compete for executive office. Most formal models of political competition, however, focus on individual decision makers: Downs de- scribed parties as \teams of politicians", but in this canonical model of electoral competition the parties are unitary actors; the principal-agent literature that studies executive accountability has few well developed models of competition, and even then the focus is on individual agents. As a consequence our understanding is limited to cases where competition is between rival leaders (as in presidential elections or primaries) or candidates (as in constituency run-offs); whilst competition between rival teams, such as that between an incumbent cabinet and its shadow, has been ignored. A critical aspect of decision-making in teams that has been commented upon in the political science literature, is the advantage faced by a principal who learns about the competencies of a set of agents who perform similar tasks. Huber and Martinez-Gallardo (2003, 2008) study executive turnover in parliamentary democracies and relate empirical patterns in the data to such a learning process: \Since party leaders will often be uncertain as to which politicians have the technical expertise and skills necessary to do their jobs well, a process of trial and error occurs to discover the best talent. This can only occur by getting rid of some ministers, bringing in new faces, and reshuffling individuals from one post to another." Similarly, Berlinski, Dewan, and Dowding (2010) show that patterns in the data on ministerial turnover in British government from 1945-2007 are related to policy failures that \provide infor- mation to the Prime Minister that was not available when she appointed her minister." Using the framework developed by Holstrom (1982), they show that the data is consistent with what we would expect to observe were a principal (the Prime Minister) to use a retention rule that reflects comparative evaluation of cabinet ministers' performance on correlated tasks These empirical studies suggest that acquisition of information about politicians' competencies is central to accountability in cabinets. However, it remains to be shown that cabinet government is an effective and efficient means for learning about such competencies when compared with other executive forms. We provide a theoretical framework that allows us to explore the cabinet per- formance against an efficient level of learning. We contrast the performance of an executive team (with respect to learning) to one that is appointed, explore the role played by competitive elections between rival (potential) executive teams, and compare outcomes under cabinet government with those that occur when decisions are taken by an individual. 3 In our model an agent is characterized by his degree of competence{either high or low{which we define as the ability to implement a policy that ex-post is in the public's best interest. The agent's type is unknown, even to himself, and is revealed only ex-post when implementing a particular type of policy. Specifically we characterize some policies as being \safe" in that the outcome is known, whereas a \reform" policy yields a high payoff only when implemented by a competent agent. As a benchmark we analyze an executive team that consists of agents who care only about the policy outcome. The first best case illustrates perfectly the logic of learning: the executive team implements the risky policy for a wide range of the parameter space even when their individual expected first period payoffs are less than those when taking the safe option. Doing so allows the team to learn the individual competencies of its members. By this trial and error process, those who have the competence necessary to do their jobs maintain their positions, whilst others do not. Indeed we find that there is more learning in larger teams. Focussing on what is individually rational, rather than what is optimal for the team, however, we find that learning is lower than in the first best case and is decreasing in the size of the team. The problem is the classic one of free-riding: an individual agent does not internalize the positive effect of his action on the team. Can the institutional framework of a cabinet alleviate this problem? Two aspects of cabinet gover- nance suggest that it should. Firstly, cabinets are (typically) formed by politicians who obtain rents from holding office and so these individual incentives should alleviate free-riding. Secondly, cabinets operate collective responsibility: the policy executed by an individual politician is chosen by the executive team under the \unanimity principle" that all government office-holders must support the same policy. This principle ensures that each politician is decisive in determining the policy to be implemented and so mitigates free-riding. We analyze a situation where prior to the execution of policy in each period, a cabinet of politicians must collectively agree whether the reform policy should be implemented. Despite the absence of free-riding, we find that learning is lower than in the first best case and is decreasing in the size of the team. In fact our results are qualitatively similar to those obtained without a cabinet. We relate this finding to the classic hold-up problem. A politician who is revealed as incompetent will vote against the further implementation of the reform policy. Anticipating this, the first period value of investment in the risky policy falls for all politicians, the Cabinet is less likely to take that course of action, and this effect grows larger in the size of the Cabinet. A hold-up occurs because a politician does not realize the full return on his investment due to the bargaining power of other politicians who can veto policy. 4 Of course there exists a resolution to this problem. Under the doctrine of individual ministerial responsibility the party, or its leader, can replace a politician who is revealed as incompetent. This should alleviate the hold-up problem since it takes away the bargaining power of any individual politician. Surprisingly, however, we show that it destroys all learning. Whilst our basic model casts doubt on the efficacy of the cabinet as a means of acquiring relevant information about politicians' competencies, it abstracts from elective and competitive elements of cabinet governance. According to both Dicey and Mill (in Representative Government), the normative force of collective responsibility is derived not from what is efficient from the cabinets's perspective but from the will of the majority. According to Marshall (1984), the interpretation of these scholars was that the convention be designed so as to \secure that the will of the majority of the electorate be put into effect.” Moreover, Mill and Bagehot, defended representative government as a process by which men and their ideas could be tested (against custom) in a competitive political process: this they called \government by discussion". These competitive and elective elements are central to cabinet governance. Whilst Parliament is in session, the daily dual between its front-benches provides the adversarial drama of British politics: on the front-benches on one side of the Commons sits the Cabinet and other government ministers; on the opposite side of the House sits Her Majesties loyal opposition, with the front-bench seats reserved for members of its Shadow Cabinet. Turner (1969) notes that \the Shadow Cabinet both derives from and influences the basic features of the British system of government, a two-party system in which voters choose directly between rival teams of potential Ministers."1 We build these competitive and elective aspects into our model in several stages. First we analyze a model (a version of the canonical principal-agent model of accountability) where voters select between a cabinet and a prospective rival team (a shadow cabinet).